Speaking was easy. Everything after it took time.
Voice notes help you get your thoughts down quickly. But people don't speak the way they write.
We repeat ourselves, pause to think, and sometimes change our minds halfway through a sentence. A transcript keeps all of that.
So after recording, you still have to listen again, clean up the words, find the main points, and put everything in order.
- 01Record
- 02Listen again
- 03Clean it up
- 04Find the important parts
- 05Organise everything
The recording saved you from typing, but it didn't save you from all the work that came after.
What if you could speak naturally and get something useful back without having to do all that work?
From a voice note to something you can use.
Handy is a Mac app that turns speech into clean, organised content.
You speak, Handy does the work in the middle, and you get something you can read, edit, and use.
The idea was simple: let people focus on what they want to say while Handy handles the rest.
- Speak
- Transcribe
- Clean
- Organise
- Edit
The user shouldn't have to worry about every step. They should just be able to speak and get on with their work.
A transcript isn't the same as a useful note.
As I thought more about the problem, I realised that transcription wasn't the main issue. Plenty of tools could already turn speech into text.
The real problem was what happened next.
Imagine talking about a project for five minutes. In that recording, you might share three ideas, mention two tasks, point out a problem, and make a decision.
A transcript gives you everything you said, in the order you said it. But finding those important parts can still take time.
The information is there. You just have to dig through it.
That changed how I thought about Handy. I didn't want to build another app that simply wrote down what people said. I wanted to help them turn their thoughts into something they could use.
Speak naturally. Let Handy handle the messy part.
I built Handy around three simple things.
Capture
Make it easy to start speaking.
Clean
Remove filler words and repetition without changing what the person means.
Organise
Turn the main points into something that's easier to read and use.
A clean paragraph is helpful, but sometimes people need more than that. They might need a summary, a list of ideas, or a set of tasks.
Handy needed to do more than write down speech. It needed to help people make sense of what they said.
One clear action was enough.
When someone opens Handy, I didn't want them to stop and figure out what to do.
They already know how to talk.
So I made the main action simple: Talk.
There's no long introduction or complicated setup before they can get started.
The interface stays out of the way and lets people do what they came to do.
Clean up the words without changing the meaning.
People don't always speak in clear, complete sentences. We use filler words, repeat ourselves, and sometimes take a long time to make a simple point.
For example, someone might say:
What was said
“So, um, I was thinking that maybe we should probably change the onboarding because people are getting stuck around verification...”
What Handy keeps
“We should change the onboarding. People are getting stuck around verification.”
The goal wasn't to make people sound different. It was to remove the extra words that made their thoughts harder to read.
The meaning still had to stay the same.
Sometimes, a clean paragraph isn't enough.
Even after cleaning up the words, a long block of text can still be hard to use.
Someone might talk about several things in one recording. I wanted Handy to make those things easier to find.
So it can organise the result into different sections.
Summary
A short version of what was said.
Key ideas
The main points worth remembering.
Action items
Things that need to be done.
Notes
Other useful details worth keeping.
The goal was to help people find what mattered without listening to the whole recording again.
They could then edit the result and make it their own. Handy gives them a starting point, not something they have to accept exactly as it is.
Four people tried Handy. Their feedback changed the product.
Before releasing Handy, I shared it with four people who had different ways of working: someone who already used transcription apps, a medical doctor, a designer, and a friend who used AI tools every day and was tired of typing.
I asked them to use Handy in their normal work and tell me what wasn't working.
Each person showed me a different problem.
The transcription user: “Don't throw away what I said.”
My early version sometimes removed words that mattered. Worse, there was no easy way to get back to the original. I changed this so the original transcript stays available. Handy cleans up a copy instead, so users can improve their text without losing what they first said.
The doctor: “The numbers have to come out exactly as I said them.”
For medical notes, changing a number or a medical term can change the meaning completely. I adjusted the cleanup so it keeps numbers and specialist terms as they are, while cleaning up the words around them.
The designer: “I just want it to stay out of my way.”
Having to open a separate app every time they wanted to speak interrupted their work. This pushed me to add a keyboard shortcut that works from the field they're already using. Handy's window only appears when needed.
The friend who uses AI tools every day: “It made my prompts sound like emails.”
This feedback changed the direction of Handy. The same polished sentence that worked in an email sounded stiff in a chat or an AI prompt. Handy was cleaning up the words without knowing where they were going.
I realised that good writing doesn't always have to sound formal. Sometimes, people want a professional email. Other times, they just want a casual message that sounds like them.
That was what led to one of Handy's most important features.
The same words can sound different in different places.
Think about writing an email to a colleague and sending a message to a friend.
You might want your email to sound clear and professional. But in a chat with a friend, that same tone could feel strange.
AI prompts are different too. You want your instructions to stay clear and natural, not turn into a formal email.
So I worked on making Handy adjust its cleanup based on the app you're using.
Your words become a clear message you can send to a colleague.
Your message stays casual instead of sounding too formal.
Your instructions stay natural and direct.
Handy checks which app you're working in, adjusts the text to fit, and puts the result where your cursor already is.
You don't have to copy the text or switch between apps. You can keep working where you are.
Context-aware cleanup is on by default, but users can turn it off and choose their own writing style in settings.
I wanted Handy to fit into people's work, not make them change how they work to use it.
AI does the work, but the user makes the final call.
Handy uses AI to clean up and organise what people say. But I didn't want users to feel like they had to accept whatever it produced.
The result is a draft. They can read it, change it, remove parts, add their own words, and decide what to do with it.
Keeping the original transcript available was part of this too. If Handy gets something wrong, users shouldn't have to start all over again.
Handy should make the work easier without taking control away from the person using it.
I designed Handy and built it myself.
Handy gave me the chance to work on the whole product, from the first idea to the finished app.
I decided what the product should do, designed the interface, worked out how the AI should behave, and built the Mac app with Claude Code. I also tested it, made changes based on feedback, and continued supporting it after release.
Building it myself changed how I worked.
I could try an idea, use it in the actual app, and see what felt wrong. I didn't have to wait for someone else to build it before I could test it.
Something could look good in Figma but feel awkward when I used it. Building Handy helped me spot those problems early.
- 01Design
- 02Build
- 03Test
- 04Improve
It wasn't a straight line. I kept moving between design and code as I learned what worked and what needed to change.
Claude Code helped me move from ideas to a working product faster. But I still had to make the important decisions about how Handy should work, what it should change, and what it should leave alone.
It reminded me that design doesn't end when the screens look right. The product has to work well in real life too.
Less time fixing words. More time using them.
Handy started with a simple problem. Speaking is faster than typing, but turning a recording into something useful still takes time.
Now, you can speak, get clean and organised text, review it, and use it where you're already working.
Feedback from the four early testers helped me improve how Handy handles the original transcript, keeps important details, fits into people's workflow, and adjusts its writing to different apps.
I designed, built, and released the Mac app in two months, and I continue to improve it.
There's still more to learn about how much time it saves people in their daily work. But Handy now does what I set out to build: it helps people get from saying something to being able to use it.
Good AI doesn't have to do everything for you.
While building Handy, I learned that making AI more useful isn't always about asking it to do more.
Sometimes, it's about knowing what not to change.
Don't change a number just to make a sentence sound better. Don't turn a casual message into a formal email. Don't remove the original words and leave the user with no way back.
I also learned how useful it is to build what you design. Using Handy myself helped me spot problems early, while testing it with other people showed me where my assumptions were wrong.
The goal was never to make people sound like AI.
It was to help them spend less time fixing their words and more time using them.
The thought is still yours. Handy just helps you get it where it needs to go.